Product — Verity AI | Authentication Infrastructure
Product authentication infrastructure · REST API · No hardware

The authentication layer
commerce is missing.

One endpoint. Decision in under 5 seconds. No hardware. No manual queue.

98%+
Accuracy in live production
<5s
End-to-end decision time
$425K
Annual savings, one production deployment
<1 day
API integration time. No rearchitecture.
How it works

One endpoint. Three inputs. A structured decision.

Step 01
Submit product images via API
Send product images with brand and category metadata. Standard HTTP POST.
POST /v1/authenticate
Step 02
Computer vision processes the submission
Computer vision compares stitching, hardware, and materials against authenticated references.
Processing <2s
Step 03
Structured decision returned instantly
Confidence score, decision, reasoning flags, and certificate URL returned in the response.
200 OK · decision: "authentic"
Capabilities

Four authentication surfaces. One API.

Same endpoint. Same schema. Same audit trail — listing, returns, warehouse, agentic.

Product Authentication
Image-based verification against authenticated references. Flags counterfeit hardware, materials, and stitching.
98%+ accuracy across supported categories
JPEG · PNG · WEBP · multi-image support
Structured confidence score + decision flag
Content Integrity
Detects manipulated, AI-generated, or misrepresentative listing images before they reach buyers.
AI-generated image detection
Image manipulation and splice detection
Listing image vs product mismatch flagging
Returns Verification
Compares returned items against the authenticated listing before a refund is issued.
Item-to-listing comparison via image API
Returns a match score and pass/fail decision
Full audit trail per return event
KYP for Agentic Commerce
For AI agents. Confidence score, certificate URL, reasoning flags — consumable by any LLM.
JSON response compatible with agentic workflows
Certificate URL for agent verification chains
Demonstrated in Mastercard's AI Payment Agent environment
API output

Structured output. Every field documented. No ambiguity.

POST /v1/authenticate · 200 OK
{
  "id": "auth_01j9xkp...",
  "decision": "authentic",
  "confidence": 0.97,
  "flags": [],
  "brand": "ganni",
  "category": "dress",
  "certificate_url": "https://verify.verityai.app/cert/...",
  "processed_at": "2025-09-14T08:22:31Z",
  "model_version": "cv-3.1"
}
decision
string
Binary outcome: "authentic" | "counterfeit" | "inconclusive". Downstream logic acts on this field.
confidence
float · 0.00 – 1.00
Model confidence. Set custom thresholds in your rules engine.
flags
array · string[]
Anomalies detected: hardware, stitching, typography, AI imagery. Empty on clean pass.
certificate_url
string · URL
Permanent authentication record. Embed in listings, receipts, or agent verification chains.
processed_at
ISO 8601
Timestamp for audit trail and compliance.
Use cases

The same API. Deployed differently per vertical.

The problem
Counterfeits reach buyers. Disputes follow. Trust erodes.
Manual authentication doesn't scale. Counterfeits pass review, disputes follow, refunds are issued.
  • Block counterfeits before they reach live listings
  • Issue a trust certificate that increases buyer confidence and CTR
  • Reduce authentication cost from $40K/month to under $5K/month
Apply for access →
$425K
Annual savings in one production deployment
98%
Reduction in authentication handling time
+17%
CTR lift on trust-badged listings
<1 day
Integration time for API-connected platforms
The problem
Return fraud is invisible until the refund is already issued.
Returns processed at volume. Swaps and fakes pass undetected — refunds issued on fraudulent items.
  • Compare returned item against the authenticated listing before any refund is issued
  • Returns a match score and decision your OMS acts on automatically
  • Eliminate returns inspection headcount
  • Full audit trail per return event for dispute resolution
Apply for access →
100%
Returns inspected - not sampled
<5s
Decision per return event
0
Manual reviewers required
Full
Audit trail per return for compliance
The problem
Inbound inventory is accepted on trust. Authenticity is assumed, not verified.
3PLs hold high-value inventory for clients. Without inbound auth, liability grows.
  • Authenticate inbound inventory at receipt via web app or API
  • Issue a certificate per item - transferable to your client
  • Flag discrepancies before storage - not after delivery
Apply for access →
Same day
Authentication at inbound - no hardware needed
Client-ready
Certificate per item, transferable to brand owner
API or
Web app - whichever fits your warehouse workflow
Full
Audit trail per item for liability coverage
The problem
Your fraud stack checks the transaction. Nobody checks the product.
Payment fraud is well-solved. Product fraud is not. No automated product-layer signal exists for retailers accepting resale.
  • Add a product-level trust signal alongside your existing fraud stack
  • Authenticate resale and returned inventory before it re-enters circulation
  • Support certified pre-owned programs with verified authentication records
  • Deploys in under a day alongside existing systems - no rearchitecture
Apply for access →
Additive
Sits alongside your existing fraud stack
<1 day
Integration - REST API, no custom middleware
Certified
Authentication records for certified pre-owned programs
JSON
Output consumable by any rules engine or platform
The problem
Payment decisions are made without knowing whether the product is real.
Payment processors operate on transaction signals. Verity adds a product-layer signal alongside.
  • Receive a structured authenticity signal at the product layer in real time
  • Confidence score and certificate URL consumable by any rules engine
  • Integrated into Mastercard's AI Payment Agent demonstration environment
Apply for access →
Real-time
Confidence score returned synchronously
Machine-readable
JSON output compatible with any rules engine or LLM
Mastercard
Partnered for their agentic commerce demonstration
Certificate
Immutable record for dispute resolution and underwriting
Production case study · Resale

Live in production.

Harvey Norman
Enterprise retail · Returns pilot
Harvey Norman returns workflow. Authentication before the refund is triggered.
Mastercard
Agentic commerce · AI Payment Agent
Integrated into the Mastercard AI Payment Agent.
Integration

Live in under a day. Four integration surfaces.

REST API
Direct API Integration
Integration time: <1 day
Standard HTTP POST. JSON in, JSON out. No SDK required.
E-Commerce
Shopify Integration
Integration time: Same day
Native Shopify. No custom development. Works at listing or checkout.
Web App
Verity Dashboard
Integration time: Zero
Browser tool for warehouse teams. Upload images, receive decisions. No integration required.
Agentic
AI Agent / LLM
Integration time: <1 day
Machine-readable JSON for LLMs or rules engines. Certificate URL for agent chains. Sub-5s.

One endpoint, four surfaces. The same POST /v1/authenticate powers every integration.

POST api.verityai.app/v1/authenticate
Trust & compliance

Enterprise deployment requirements, addressed.

Zero PII stored
Images processed and discarded. No personal data retained.
Govt. Backed
Australian Government backed and funded.
Audit trail
Every event logged: timestamp, model version, certificate URL.
TLS 1.3
All traffic encrypted in transit. No unencrypted endpoints.
EU AI Act
Compliant for automated decision systems in commercial contexts.
SOC 2
Type II in progress. Documentation available under NDA.
NDA ready
Standard NDA available. Security docs on request.
GDPR aligned
GDPR-aligned. DPA available on request.
Request access

Start with a scoped deployment. Live in five business days.

Every deployment is scoped first. Working proof of concept inside five business days.

  • Scoped to your specific workflow and integration surface
  • Live in under five business days from access granted
  • Full technical support throughout deployment
Apply for Access
Qualified operators only · Response within one business day
Qualified operators only · Response within one business day